1 /*
2 * Copyright 2019-2020 The OpenSSL Project Authors. All Rights Reserved.
3 *
4 * Licensed under the Apache License 2.0 (the "License"). You may not use
5 * this file except in compliance with the License. You can obtain a copy
6 * in the file LICENSE in the source distribution or at
7 * https://www.openssl.org/source/license.html
8 */
9
10 #include <stddef.h>
11 #include <string.h>
12 #include <stdio.h>
13 #include <openssl/e_os2.h>
14 #include "crypto/punycode.h"
15
16 static const unsigned int base = 36;
17 static const unsigned int tmin = 1;
18 static const unsigned int tmax = 26;
19 static const unsigned int skew = 38;
20 static const unsigned int damp = 700;
21 static const unsigned int initial_bias = 72;
22 static const unsigned int initial_n = 0x80;
23 static const unsigned int maxint = 0xFFFFFFFF;
24 static const char delimiter = '-';
25
26 #define LABEL_BUF_SIZE 512
27
28 /*-
29 * Pseudocode:
30 *
31 * function adapt(delta,numpoints,firsttime):
32 * if firsttime then let delta = delta div damp
33 * else let delta = delta div 2
34 * let delta = delta + (delta div numpoints)
35 * let k = 0
36 * while delta > ((base - tmin) * tmax) div 2 do begin
37 * let delta = delta div (base - tmin)
38 * let k = k + base
39 * end
40 * return k + (((base - tmin + 1) * delta) div (delta + skew))
41 */
42
adapt(unsigned int delta,unsigned int numpoints,unsigned int firsttime)43 static int adapt(unsigned int delta, unsigned int numpoints,
44 unsigned int firsttime)
45 {
46 unsigned int k = 0;
47
48 delta = (firsttime) ? delta / damp : delta / 2;
49 delta = delta + delta / numpoints;
50
51 while (delta > ((base - tmin) * tmax) / 2) {
52 delta = delta / (base - tmin);
53 k = k + base;
54 }
55
56 return k + (((base - tmin + 1) * delta) / (delta + skew));
57 }
58
is_basic(unsigned int a)59 static ossl_inline int is_basic(unsigned int a)
60 {
61 return (a < 0x80) ? 1 : 0;
62 }
63
64 /*-
65 * code points digit-values
66 * ------------ ----------------------
67 * 41..5A (A-Z) = 0 to 25, respectively
68 * 61..7A (a-z) = 0 to 25, respectively
69 * 30..39 (0-9) = 26 to 35, respectively
70 */
digit_decoded(const unsigned char a)71 static ossl_inline int digit_decoded(const unsigned char a)
72 {
73 if (a >= 0x41 && a <= 0x5A)
74 return a - 0x41;
75
76 if (a >= 0x61 && a <= 0x7A)
77 return a - 0x61;
78
79 if (a >= 0x30 && a <= 0x39)
80 return a - 0x30 + 26;
81
82 return -1;
83 }
84
85 /*-
86 * Pseudocode:
87 *
88 * function ossl_punycode_decode
89 * let n = initial_n
90 * let i = 0
91 * let bias = initial_bias
92 * let output = an empty string indexed from 0
93 * consume all code points before the last delimiter (if there is one)
94 * and copy them to output, fail on any non-basic code point
95 * if more than zero code points were consumed then consume one more
96 * (which will be the last delimiter)
97 * while the input is not exhausted do begin
98 * let oldi = i
99 * let w = 1
100 * for k = base to infinity in steps of base do begin
101 * consume a code point, or fail if there was none to consume
102 * let digit = the code point's digit-value, fail if it has none
103 * let i = i + digit * w, fail on overflow
104 * let t = tmin if k <= bias {+ tmin}, or
105 * tmax if k >= bias + tmax, or k - bias otherwise
106 * if digit < t then break
107 * let w = w * (base - t), fail on overflow
108 * end
109 * let bias = adapt(i - oldi, length(output) + 1, test oldi is 0?)
110 * let n = n + i div (length(output) + 1), fail on overflow
111 * let i = i mod (length(output) + 1)
112 * {if n is a basic code point then fail}
113 * insert n into output at position i
114 * increment i
115 * end
116 */
117
ossl_punycode_decode(const char * pEncoded,const size_t enc_len,unsigned int * pDecoded,unsigned int * pout_length)118 int ossl_punycode_decode(const char *pEncoded, const size_t enc_len,
119 unsigned int *pDecoded, unsigned int *pout_length)
120 {
121 unsigned int n = initial_n;
122 unsigned int i = 0;
123 unsigned int bias = initial_bias;
124 size_t processed_in = 0, written_out = 0;
125 unsigned int max_out = *pout_length;
126
127 unsigned int basic_count = 0;
128 unsigned int loop;
129
130 for (loop = 0; loop < enc_len; loop++) {
131 if (pEncoded[loop] == delimiter)
132 basic_count = loop;
133 }
134
135 if (basic_count > 0) {
136 if (basic_count > max_out)
137 return 0;
138
139 for (loop = 0; loop < basic_count; loop++) {
140 if (is_basic(pEncoded[loop]) == 0)
141 return 0;
142
143 pDecoded[loop] = pEncoded[loop];
144 written_out++;
145 }
146 processed_in = basic_count + 1;
147 }
148
149 for (loop = processed_in; loop < enc_len;) {
150 unsigned int oldi = i;
151 unsigned int w = 1;
152 unsigned int k, t;
153 int digit;
154
155 for (k = base;; k += base) {
156 if (loop >= enc_len)
157 return 0;
158
159 digit = digit_decoded(pEncoded[loop]);
160 loop++;
161
162 if (digit < 0)
163 return 0;
164 if ((unsigned int)digit > (maxint - i) / w)
165 return 0;
166
167 i = i + digit * w;
168 t = (k <= bias) ? tmin : (k >= bias + tmax) ? tmax : k - bias;
169
170 if ((unsigned int)digit < t)
171 break;
172
173 if (w > maxint / (base - t))
174 return 0;
175 w = w * (base - t);
176 }
177
178 bias = adapt(i - oldi, written_out + 1, (oldi == 0));
179 if (i / (written_out + 1) > maxint - n)
180 return 0;
181 n = n + i / (written_out + 1);
182 i %= (written_out + 1);
183
184 if (written_out > max_out)
185 return 0;
186
187 memmove(pDecoded + i + 1, pDecoded + i,
188 (written_out - i) * sizeof *pDecoded);
189 pDecoded[i] = n;
190 i++;
191 written_out++;
192 }
193
194 *pout_length = written_out;
195 return 1;
196 }
197
198 /*
199 * Encode a code point using UTF-8
200 * return number of bytes on success, 0 on failure
201 * (also produces U+FFFD, which uses 3 bytes on failure)
202 */
codepoint2utf8(unsigned char * out,unsigned long utf)203 static int codepoint2utf8(unsigned char *out, unsigned long utf)
204 {
205 if (utf <= 0x7F) {
206 /* Plain ASCII */
207 out[0] = (unsigned char)utf;
208 out[1] = 0;
209 return 1;
210 } else if (utf <= 0x07FF) {
211 /* 2-byte unicode */
212 out[0] = (unsigned char)(((utf >> 6) & 0x1F) | 0xC0);
213 out[1] = (unsigned char)(((utf >> 0) & 0x3F) | 0x80);
214 out[2] = 0;
215 return 2;
216 } else if (utf <= 0xFFFF) {
217 /* 3-byte unicode */
218 out[0] = (unsigned char)(((utf >> 12) & 0x0F) | 0xE0);
219 out[1] = (unsigned char)(((utf >> 6) & 0x3F) | 0x80);
220 out[2] = (unsigned char)(((utf >> 0) & 0x3F) | 0x80);
221 out[3] = 0;
222 return 3;
223 } else if (utf <= 0x10FFFF) {
224 /* 4-byte unicode */
225 out[0] = (unsigned char)(((utf >> 18) & 0x07) | 0xF0);
226 out[1] = (unsigned char)(((utf >> 12) & 0x3F) | 0x80);
227 out[2] = (unsigned char)(((utf >> 6) & 0x3F) | 0x80);
228 out[3] = (unsigned char)(((utf >> 0) & 0x3F) | 0x80);
229 out[4] = 0;
230 return 4;
231 } else {
232 /* error - use replacement character */
233 out[0] = (unsigned char)0xEF;
234 out[1] = (unsigned char)0xBF;
235 out[2] = (unsigned char)0xBD;
236 out[3] = 0;
237 return 0;
238 }
239 }
240
241 /*-
242 * Return values:
243 * 1 - ok, *outlen contains valid buf length
244 * 0 - ok but buf was too short, *outlen contains valid buf length
245 * -1 - bad string passed
246 */
247
ossl_a2ulabel(const char * in,char * out,size_t * outlen)248 int ossl_a2ulabel(const char *in, char *out, size_t *outlen)
249 {
250 /*-
251 * Domain name has some parts consisting of ASCII chars joined with dot.
252 * If a part is shorter than 5 chars, it becomes U-label as is.
253 * If it does not start with xn--, it becomes U-label as is.
254 * Otherwise we try to decode it.
255 */
256 char *outptr = out;
257 const char *inptr = in;
258 size_t size = 0;
259 int result = 1;
260
261 unsigned int buf[LABEL_BUF_SIZE]; /* It's a hostname */
262 if (out == NULL)
263 result = 0;
264
265 while (1) {
266 char *tmpptr = strchr(inptr, '.');
267 size_t delta = (tmpptr) ? (size_t)(tmpptr - inptr) : strlen(inptr);
268
269 if (strncmp(inptr, "xn--", 4) != 0) {
270 size += delta + 1;
271
272 if (size >= *outlen - 1)
273 result = 0;
274
275 if (result > 0) {
276 memcpy(outptr, inptr, delta + 1);
277 outptr += delta + 1;
278 }
279 } else {
280 unsigned int bufsize = LABEL_BUF_SIZE;
281 unsigned int i;
282
283 if (ossl_punycode_decode(inptr + 4, delta - 4, buf, &bufsize) <= 0)
284 return -1;
285
286 for (i = 0; i < bufsize; i++) {
287 unsigned char seed[6];
288 size_t utfsize = codepoint2utf8(seed, buf[i]);
289 if (utfsize == 0)
290 return -1;
291
292 size += utfsize;
293 if (size >= *outlen - 1)
294 result = 0;
295
296 if (result > 0) {
297 memcpy(outptr, seed, utfsize);
298 outptr += utfsize;
299 }
300 }
301
302 if (tmpptr != NULL) {
303 *outptr = '.';
304 outptr++;
305 size++;
306 if (size >= *outlen - 1)
307 result = 0;
308 }
309 }
310
311 if (tmpptr == NULL)
312 break;
313
314 inptr = tmpptr + 1;
315 }
316
317 return result;
318 }
319
320 /*-
321 * a MUST be A-label
322 * u MUST be U-label
323 * Returns 0 if compared values are equal
324 * 1 if not
325 * -1 in case of errors
326 */
327
ossl_a2ucompare(const char * a,const char * u)328 int ossl_a2ucompare(const char *a, const char *u)
329 {
330 char a_ulabel[LABEL_BUF_SIZE];
331 size_t a_size = sizeof(a_ulabel);
332
333 if (ossl_a2ulabel(a, a_ulabel, &a_size) <= 0) {
334 return -1;
335 }
336
337 return (strcmp(a_ulabel, u) == 0) ? 0 : 1;
338 }
339